"As an individual, I suppose you can't really do much": Environmental Sustainability Perceptions of Machine Learning Practitioners

TY - CHAP

T1 - "As an individual, I suppose you can't really do much"

T2 - Environmental Sustainability Perceptions of Machine Learning Practitioners

AU - Görücü, Sinem

AU - Ren, Yuheng

AU - Samuel, Gabrielle

AU - Panagiotidou, Georgia

N1 - Publisher Copyright: © 2025 Copyright held by the owner/author(s).

PY - 2025/6/23

Y1 - 2025/6/23

N2 - The growing need for computational resources and energy to develop, train, and run Machine Learning models has been gaining increasing attention. Still, while a considerable body of research is focusing on tools and methods to calculate the energy consumption and carbon footprint of ML processes, there is a distinctive lack of research into the perceptions, practices, challenges and behaviors of ML practitioners regarding the environmental sustainability of their work. To address this gap, we conducted a series of semi-structured interviews with ML practitioners from academia and industry to understand how they engage with the topic within their teams. We found that sustainability, even within our relatively motivated interview cohort, was still seen as a gnice-To-have' with considerations such as speed, market competition, or time management considered as having higher priority. We discuss these findings in-light of prior work on sustainability within household and work settings and highlight the need for individual as well as collective responsibility and agency. This work contributes to the design and development of future ML eco-feedback tools and promotes sustainable awareness and practices within this context.

AB - The growing need for computational resources and energy to develop, train, and run Machine Learning models has been gaining increasing attention. Still, while a considerable body of research is focusing on tools and methods to calculate the energy consumption and carbon footprint of ML processes, there is a distinctive lack of research into the perceptions, practices, challenges and behaviors of ML practitioners regarding the environmental sustainability of their work. To address this gap, we conducted a series of semi-structured interviews with ML practitioners from academia and industry to understand how they engage with the topic within their teams. We found that sustainability, even within our relatively motivated interview cohort, was still seen as a gnice-To-have' with considerations such as speed, market competition, or time management considered as having higher priority. We discuss these findings in-light of prior work on sustainability within household and work settings and highlight the need for individual as well as collective responsibility and agency. This work contributes to the design and development of future ML eco-feedback tools and promotes sustainable awareness and practices within this context.

UR - https://www.scopus.com/pages/publications/105010813792

U2 - 10.1145/3715275.3732088

DO - 10.1145/3715275.3732088

M3 - Conference paper

T3 - ACMF AccT 2025 - Proceedings of the 2025 ACM Conference on Fairness, Accountability,and Transparency

SP - 1312

EP - 1324

BT - ACMF AccT 2025 - Proceedings of the 2025 ACM Conference on Fairness, Accountability,and Transparency

PB - Association for Computing Machinery (ACM)

ER -